On Mon, November 16, 2015 6:09 pm, Mark Connor wrote:

I have approximately 500 Linux servers managed my xymon. My managed clients reboot monthly. A small percentage come up purple and stay purple until the client software is restarted. Sound familiar?

Hi Mark,

Short answer is that that shouldn't happen. Overall, the client is designed to be pretty fail safe. Assuming that the client is configured to start on boot, that's an indication that there's something unusual going on. Going purple is just by design there (purple = stale alert; something that was sending statuses hasn't sent something recently).

On a machine where this has happened, are there any processes stuck at all? Typically, xymonlaunch should be the only process that remains running, aside from backgrounded 'vmstat's. Everything else will be a transient shell script or the 'xymon' binary.

Assuming xymonlaunch IS running, try sending it a -USR2 signal to toggle debug mode. Its log file should indicate if it's having problems executing something or if there's something odd about its environment. (cat /proc/<pidoflxymonlaunch>/environ to examine.)

If nothing seems obvious there but nothing is making it in, edit xymonclient.sh to add "--debug" to the $XYMON execution call near the bottom ($XYMON --debug $XYMSRV ...), before restarting it, and verify in the log file that it's sending to where it's supposed to be.

HTH, -jc